Autism Spectrum Screening Questionnaire (ASSQ) in a School-Aged Population in Korea

Background: Effective and efficient screeners play a critical role in early identification and intervention of children with ASD. The 27-item Autism Spectrum Screening Questionnaire (ASSQ) includes items that assess core ASD features (e.g., social-communication and restricted, repetitive behaviors), as well as other behaviors (e.g., motor clumsiness). The ability of the ASSQ to distinguish autism from non-autism diagnoses is well established in European children (see Ehlers et al., 1999, Posserud et al., 2009, and Mattila et al., 2009).

Objectives: Evaluate psychometric properties of the ASSQ-Korean version (K-ASSQ) in children with ASD drawn from a Korean population-based sample.

Methods: Participants were drawn from two populations: (1) Epidemiological sample including all 7-12-year-old children in a target Korean city (N=24,985, mean ± standard deviation age: 9.08±2.23 years) who were screened with parent- and/or teacher ASSQ (see Kim et al, 2011). For screen positive children (parental ASSQ scores in upper 5^thpercentile and/or teacher ASSQ scores ≥ 10; n=296), best-estimate clinical diagnoses of DSM-IV PDD, DSM-5 ASD and Social Communication Disorder (SCD) were generated, based on ADI-R, ADOS and cognitive assessment,  (2) Clinical sample including 135 (80% male) 7-12-year-old children (8.74±1.84 years) who were evaluated at a child and adolescent psychiatric clinic in four university/university-affiliated medical centers. Diagnoses included ADHD (67.4%), Disruptive behavioral disorders (ODD+CD: 7.4%), Anxiety Disorders (15.6%), Depressive Disorders (11.1%), and others (22.2%), with 27.7% having multiple diagnoses.

Convergent validity was examined with the Social Responsiveness Scale (SRS; Constantino & Gruber, 2005). Principal component analyses with varimax rotation were conducted to assess ASSQ factor structure.  Nonparametric Mann–Whitney U test was used to compare ASSQ scores of children with ASD to non-autism diagnoses.  Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of the ASSQ were computed.

Results: The correlation between the parent ASSQ and Social Responsiveness Scale (SRS) scores was moderate (0.3 < |correlation coefficient| < 0.7) in the overall epidemiological and clinical samples. ASSQ scores were not significantly associated with IQ. Factor analysis yielded two factors; the first factor included items measuring social interaction, whereas the second included items related to repetitive behaviors, being consistent with revised DSM5 criteria for ASD. Parent ASSQ scores differentiated children with ASD from unaffected children in the epidemiological sample (23.65±9.68 (n=172) vs. 4.39±5.13 (n=23,138); for DSM-5 ASD). In the clinical sample, children with ADHD had higher ASSQ scores than children with affective or disruptive behavior disorders (10.44±8.30 (n=68) vs. 6.87±6.07 (n=15)), but the difference did not reach statistical significance (p-value = 0.176). Using the proposed European cut-off of 13 (Ehlers et al., 1999), sensitivity was high (90.9; PPV=62.0) but specificity was poor (9.3; NPV=38.5). Using cut-offs of ≥ 14 (sensitivity=88.6; specificity=10.2) or ≥ 15 (sensitivity=84.7; specificity=15.7) did not yield considerable improvement.

Conclusions: The present study suggests that the ASSQ can be employed as a valid and effective screening instrument for identification of children with ASDs in Korean children, with a cut-off of 13 for parent scores.